GPT-5.5 Instant
OpenAI's fast ChatGPT model from May 2026, no longer the ChatGPT default
Context window
Not published
Input / 1M tokens
ChatGPT only (no API)
Output / 1M tokens
ChatGPT only (no API)
Provider
OpenAI
GPT-5.5 Instant had no model ID of its own in the API. The chat-latest alias, which tracks the current ChatGPT Instant model, costs $5 input / $0.50 cached / $30 output per 1M and now points to a newer model (knowledge cutoff April 20, 2026). · Data verified 2026-11-08
GPT-5.5 Instant was OpenAI's fast 'Instant' tier and became ChatGPT's default model on May 5, 2026, replacing GPT-5.3 Instant. It was tuned for speed and everyday interactive use rather than maximum reasoning depth, accepted text and image input, and had a knowledge cutoff of August 2025. It never got its own API model ID; the API's chat-latest alias tracks whichever Instant model ChatGPT currently uses. In August 2026 OpenAI replaced it as the default with GPT-5.6 Luna for Free users and GPT-5.6 Sol for Plus and Pro, according to Business Today, and ChatGPT now runs GPT-6 Luna and GPT-6 Sol.
Capability index
Relative estimates (0-100) to place this model against its peers, grounded in published benchmarks.
How to access it
No longer offered as the ChatGPT default: OpenAI moved ChatGPT to GPT-5.6 models in August 2026 and since then to GPT-6 Luna (Free and Go) and GPT-6 Sol (paid plans). For API work, use chat-latest for the current ChatGPT Instant model or a GPT-6 model such as gpt-6-luna.
Strengths
- ✓Low-latency, snappy responses for interactive and chat use
- ✓Multimodal text + image input
- ✓Large 400K-token context window for the fast tier
- ✓Sensible default for everyday ChatGPT tasks
Best for developers who...
When to choose it (and when not to)
Reach for GPT-5.5 Instant when...
- →Interactive chat and everyday assistant tasks where responsiveness matters most
- →High-volume, latency-sensitive workloads that do not need deep reasoning
- →General questions, drafting, and summarization at speed
Look elsewhere if...
- ✕Hard reasoning, math, or complex agentic coding (use GPT-6.1 Sol or GPT-6 Astra)
- ✕Any API integration: it has no model ID of its own (use chat-latest or gpt-6-luna)
- ✕New work: ChatGPT no longer uses it as the default model
How to use it
- ›Keep prompts concise and direct - Instant is tuned for quick turnaround, not extended deliberation
- ›For multi-step reasoning or long agentic tasks, switch to the full GPT-5.5 model
- ›Provide needed context inline since it favors fast single-pass answers
Quickstart
NoteGPT-5.5 Instant was a ChatGPT model with no API model id of its own. For fast, low-cost API calls use GPT-6 Luna (gpt-6-luna).ChatGPT replaced it as the default model in August 2026.
Benchmarks
| Benchmark | Score | Notes |
|---|---|---|
| AIME 2025 | 81.2 | GPT-5.5 Instant internal evaluation |
| Terminal-Bench 2.0 | 82.7 | OpenAI reported benchmark |
| FrontierMath Tier 1 - 3 | 51.7 | OpenAI reported benchmark |
Compare GPT-5.5 Instant
Compare GPT-5.5 Instant with any other model
Build a comparison →All model comparisons →